Many medical procedures rely on imaging for guidance of procedures, particularly those that are minimally invasive such as needle procedures. Needle procedures are routinely performed to deliver drugs, take tissue samples, or perform therapy. Therapies may include, but are not limited to, tissue ablation therapies such as radiofrequency ablation, cryoablation, photodynamic therapy, brachytherapy, radiation, laser and microwave ablation; implant of a device such as an artificial heart valve, stent, stent graft, feeding tube, catheter, radioactive seed or electrode; establishment of a channel or pathway such as a shunt; bypass or closure or surgical resection of a portion of tissue; or to place a localization marker or fiducial that can then guide subsequent surgery, radiation therapy etc. to the appropriate location. Many other minimally invasive therapies exist.
When performing these and other minimally invasive interventional procedures, it is important that a physician know the position and orientation of surgical instruments relative to the tissue of interest. While this is sometimes obvious (e.g., direct visualization of an obviously differentiated tissue type), it is often not. Sometimes, diseased tissue may not look different than normal surrounding tissue. Sometimes, an instrument's tip may not be directly visualized and, occasionally, the tissue may not be directly visualized at all. This is especially true for minimally invasive procedures where it is desirable to create as small an entry as possible so structures and tissues of interest may never be visualized.
In many cases, these procedures may be carried out with the assistance of volumetric imaging such as Computed Tomography (CT), Magnetic Resonance Imaging (MRI), or Positron Emission Tomography (PET). They may also be carried out using optical techniques including direct visualization through an endoscope, or through the use of some kind of spectroscopy or fluorescence. Ultrasound imaging (US) and X-ray imaging are also used extensively.
Imaging modalities may also be used before an interventional procedure to plan the treatment or diagnostic procedure, or during the interventional procedure itself to help locate the tissue and/or instruments. X-ray, optical imaging, and US are often regarded as real-time imaging modalities because they may be more portable and convenient than volumetric modalities, and can be easily used during an intervention. In some cases, these imaging devices may not offer as much information as volumetric modalities such as MRI. For example, certain tumors or anatomy that are visible on MRI may not be apparent on US or X-ray, or the quality may be insufficient.
Currently, accurate and easy targeting of a biopsy or therapy device into a target site seen under volumetric methods once a patient has been moved out of the scanner, particularly a CT or MRI scanner, is a challenge since the live device position is not seen on the scans once the scans are complete. The same is true for Cone Beam Computed Tomography (CBCT) and other modalities. While it is possible to perform an intervention in the scanner itself, this may be time consuming, inconvenient, and costly. Ultimately, while many minimally invasive interventions such as needle procedures do have pre-procedure volumetric imaging available, the procedure itself is performed with the assistance of rudimentary imaging devices such as US or X-ray with the pre-procedure volumetric images available only as static films or on a display workstation.
In some cases, such as those performed under X-ray, the anatomy may be visible only during injection of contrast and while the X-ray beam is on. This may expose the surgical team and patient to frequent doses of ionizing radiation and the patient to high doses of nephrotoxic contrast agents. Standard X-rays offer only two dimensional views requiring frequent repositioning of the imager to ensure the instrument is in the correct 3D location in the anatomy.
In some instances, such as those performed using ultrasound (or modifications of US such as Contrast Enhanced Ultrasound (CEUS), or ultrasound elastography), the anatomy may be poorly visualized or presented in a form that makes it difficult for a physician to interpret. Some lesions or anatomy may not be suited to ultrasound at all so it is necessary to “mentally fuse” images from preoperatively-obtained volumetric images with the live ultrasound images.
A physician may also have difficulty identifying a target on the live modality that was previously seen on the preoperative images especially if they are of different modalities. In some cases, the lesion may be completely invisible under a live modality.
Additionally, it is often desirable to perform a minimally invasive procedure to minimize chances of severe complications that sometimes accompany surgery. By precisely targeting devices, focal cancer lesions treatments, therapy or biopsy locations, it may be possible without subjecting the patient to a large, invasive procedure that would otherwise be poorly tolerated. Only the diseased tissue may be targeted, and healthy tissue spared.
In most cases, the location of an instrument or device must be precisely known in order to properly treat a patient. For example, the location and orientation of a heart valve delivered by a transapical approach must be known prior to deployment. Other examples include placement of biopsy needles prior to sampling, placement of therapy devices as listed above or devices such as implanted fiducials for marking of tumor boundaries for use in later surgery or radiation therapy. In some cases, needles or other instruments may be inserted to monitor therapy. For example, temperature sensors in the form of needles may be inserted to monitor an ablation procedure. When a plurality of devices are implanted such as needles designed to sample multiple locations or ablate multiple portions of a tumor, it is important to accurately place each in a desired location to ensure the therapy is correctly administered.
Currently, needles and other minimally invasive devices may be directed to targets using techniques that may include freehand placement of needles, freehand image-guided needle procedures, needle guides, transperineal saturation guidance, stereotactic frames, robots, or computer assisted image guided intervention.
Needle procedures may be performed freehand and without the use of imaging if the target is large or apparent, such as a large palpable lump or nodule, however image guidance is often preferred. During the technique of freehand needle procedures, needles are typically held in a physician's hand and inserted into the lesion of interest.
Image-guided freehand procedures are similar except that, from time-to-time during the insertion process (or continuously in some cases), an X-ray, CT scan, US, MRI scan, etc. is used to ensure the needle is properly approaching the target and is not impinging on sensitive structures. This is a very common type of needle procedure. For example, during freehand ultrasound-guided needle procedures, an ultrasound transducer may be used to visualize the lesion and path. A needle is then introduced within the scan plane of the transducer so that it can be visualized on its path toward the lesion. This approach may be difficult if the target cannot be easily identified, and may be time consuming or use copious amounts of radiation or contrast agents if X-ray imaging is used.
Another common approach uses “needle guides” that are employed during some ultrasound procedures. In this case, a special guide tube may be attached to an ultrasound transducer. This needle guide is positioned in a known orientation and location relative-to and in the scan plane of the probe, usually by a “click-on” alignment feature. Once attached to the transducer, the paths of a needle placed into the guidance portion of the needle guide can be predicted along the specific path predefined by the guide. The needle path is displayed on the ultrasound screen as a fixed line, and this path is aligned with the lesion and the needle placed into the target for biopsy, treatment, etc. An example of a needle guide may be found in U.S. Pat. No. 8,073,529 to Cermak et al. which is hereby incorporated by reference herein in its entirety.
Prostate biopsy sometimes makes use of a Transrectal Ultrasound (TRUS). A TRUS probe may be positioned in a patient's rectum and a needle guide may be attached to the probe. The tube on the needle guide is used to direct needles into lesions that can be visualized on the ultrasound or, more often, to ensure the needles are sampling within the prostate and not elsewhere.
Needle guides are typically useful in cases where a target is readily visualized by ultrasound, and can restrict the approach (of a physician to the target) to an approach that is in-plane with the viewing plane of the ultrasound. In the prostate, some experts have indicated that transrectal approaches may lead to a greater incidence of infection, and transperineal biopsies may be superior for prostate needle procedures. Needle guides would likely not be useful for Transperineal biopsies.
An alternate approach is a technique known as transperineal saturation biopsy, a template consisting grid of regularly spaced parallel holes placed externally adjacent to the perineum of the patient. A transrectal ultrasound may be introduced and used to observe the sequential placement of needles through the holes in the grid. Needles are inserted into each hole that covers part of the prostate in succession and a sample of the tissue is taken.
Saturation biopsies can be expensive and time-consuming due to the large number of samples (e.g., typically at least sixty, and sometimes twice that number) that are extracted and analyzed. They may also be uncomfortable for the patient.
In some cases, partial or “focal” transperineal biopsies may be performed in which a subset of a saturation biopsy is used to selectively target certain locations within the prostate or tissue being sampled. Based on a scan or other knowledge of the probable location of the cancer (such as the results of a prior biopsy), the suspected area may be preferentially sampled. Even in these reduced biopsies, usually at least 30 biopsy cores are obtained.
Various robotically-assisted biopsy techniques are known, using multi-axis robots that serve as a needle aiming and holding devices. Based on preoperative volumetric scans, a robot is first registered to a patient. The needle held by the robot is then aligned to the target automatically. A physician delivers the needle to the target either by hand pushing the needle or by directing the robot to do so using an electromechanical control mechanism.
Stereotactic biopsy has been used for many years. In this method, a frame is fitted to the patient, typically to the head, in order to obtain needle access to a lesion in the brain. The location of the target relative to the location of the frame is determined from scans, and a needle on the frame is aligned to the target using dials and precise scales to move and angle the needle. It is then inserted in a straight path through a trephination or burr hole into the location in the brain.
Various forms of stereotaxy exist, but the technique is currently mainly limited to radiosurgery or radiotherapy using external radiation beams as stereotactic frames, and needles are rarely used any more. The technique is regarded as complex and fairly invasive, and has been largely replaced by computer assisted “frameless stereotaxy” (described below).
The advent of accurate and inexpensive position sensors has enabled methods of Image-Guided Interventions (IGI) (also known as “frameless stereotaxy”) to be used to bring an instrument to the location of a target during an interventional procedure. Proper localization including position and orientation of these devices is critical to obtain the best result and patient outcome.
Some IGI systems use an externally placed locating device (also known as tracking systems or position sensors), such as a camera system or magnetic field generator together with an instrument containing a trackable component or “position indicating element” that can be localized by a locating device or tracking system (collectively referred to hereinafter a “tracking device”). Depending on the device and technology, these use infrared light emitting diodes (LEDs), reflective spheres, or small electromagnetic sensing coils as position indicating elements.
Position indicating elements are associated with a coordinate system and are typically attached to instruments such as surgical probes, drills, microscopes, needles, X-ray machines, etc., and to the patient. The spatial coordinates and often the orientation (depending on the technology used) of the coordinate system associated with the position indicating elements can be determined by the tracking device in the fixed coordinate system (or fixed “frame of reference”) of the tracking device. Many tracking devices may be able to track multiple position indicating elements simultaneously in their fixed frame of reference. Through geometrical transformations, it is possible to determine the position and orientation of any position indicating element relative to a frame of reference of any other position indicating element tracked by the same tracking device.
A variety of different tracking devices exist, having different advantages and disadvantages. For example, optical tracking devices may be constructed to enable the highly accurate position and orientation of a tool equipped with position indicating elements to be calculated. An example of an optical tracking device is the Polaris Vicra (Northern Digital Inc., Waterloo, ON Canada). Optical tracking devices suffer from line-of-site constraints, as they rely on triangulation of a light-emitting diode or reflective marker with several cameras.
An example of an Electromagnetic (EM) tracking device is the Aurora (Northern Digital Inc., Waterloo, ON Canada). EM tracking devices do not require a line-of-sight between the tracking device and the position indicating elements. EM tracking devices may be used with flexible instruments where position indicating elements are placed at the tip of the instruments. Other known tracking devices include, but are not limited to, mechanical linkage devices, fiber optic devices, ultrasonic devices and global positioning devices.
Image guided interventions using these systems can be effectively performed if an accurate “registration” is available to mathematically map the position data of position indicating elements expressed in terms of the coordinate system of the tracking device (“patient space”) to the coordinate system of the externally imaged data (“image space”) determined at the time the images were taken. In rigid objects such as the skull or bones, one method of registration is performed by using a probe equipped with position indicating elements (therefore, the probe itself is tracked by a tracking device) to touch fiducial markers (such as, for example, small steel balls (x-spots) made by the Beekley Corporation, Bristol, Conn.) placed on the patient prior to imaging. This enables the system to obtain the patient space coordinates of the fiducials. These same fiducials are visible on an image such as, for example, a CT scan and are identified in the image space by indicating them, for example, on a computer display. Once these same markers are identified in both spaces, a registration transformation or equivalent mathematical construction can be calculated. In one commonly used form, a registration transformation may comprise a 4×4 matrix that embodies the translations, magnification factors and rotations required to bring the markers (and thus the coordinate systems) in one space in to coincidence with the same markers in the other space.
Fiducial markers used for registration may be applied to objects such as bone screws or stick-on markers that are visible to the selected imaging device, or can be implicit, such as unambiguous parts of the patient anatomy. These anatomical fiducials may include unusually shaped bones, osteophytes or other bony prominence, calcifications, features on blood vessels or other natural lumens (such as bifurcations of bronchial airways), individual sulci of the brain, or other markers that can be unambiguously identified in the image and patient. A rigid affine transformation such as the 4×4 matrix described above may require the identification of at least three pairs of non-collinear points in the image space and the patient space. Often, many more points are used and a best-fit may be used to optimize the registration. It is normally desirable that fiducials remain fixed relative to the anatomy from the time of imaging until the time that registration is complete.
Registration for image-guided surgery may be accomplished using different methods. Paired-point registration (described above) is accomplished by a user identifying points in image space and then obtaining the coordinates of the corresponding points in patient space.
Another type of registration, surface registration, can be done in combination with, or independent of, paired point registration. In surface registration, a cloud of points is obtained in the patient space and matched with a surface model of the same region in image space. A best-fit transformation relating one surface to the other may then be calculated. In another type of registration, repeat-fixation devices may be used that involve a user repeatedly removing and replacing a device in known relation to the patient or image fiducials of the patient.
Automatic registration may also be performed. Automatic registration may, for example, make use of predefined fiducial arrays or “fiducial shapes” that are readily identifiable in image space by a computer. The patient space position and orientation of these arrays may be inferred through the use of a position indicating element fixed to the fiducial array. Other registration methods also exist, including methods that attempt to register non-rigid objects generally through image processing means.
Registrations may also be performed to calculate transformations between separately acquired images. This may performed by identifying “mutual information” (e.g., the same fiducial markers existing in each image set). In this regard, information visible in one image, but not the other, may be coalesced into a combined image containing information from both.
One such method for doing “image-image co-registration” for ultrasound and MRI was presented by Xu et al in “Real-time MRI-TRUS Fusion for Guidance of Targeted Prostate Biopsies,” Computer Aided Surg., 2008 September; 13(5): 255-264. Another method of registration of pre-procedure and intra-procedure images is disclosed in U.S. patent application Ser. No. 13/918,413 to Glossop et al., filed Jun. 14, 2013, entitled “System, Method and Device for Prostate Diagnosis and Intervention,” each of which is hereby incorporated by reference herein in its entirety. These methods include the co-registration or matching of two sets of similar but non-identical three dimensional images. The images are not identical even when the same modality is used due to the movement of tissue and the patient between the times of the scans. When the modalities differ (e.g., ultrasound and MRI), the images also differ. Co-registration may take the form of rigid, affine, non-rigid (deformable) etc. methods, many of which are well known in the art and are a continuous area of research.
Once the images have been co-registered, a mapping is available that is able to take a point or region on one image set and transfer it to the other image set.
In certain implementations, the location of lesions, targets or regions of interest may be copied or transferred on to other images. For example, if a region or target was detected on MRI, it may be transferred onto CT images, X-ray images, PET images, Ultrasound images, or other MRI images, for example. This may be done, for instance, by using the aforementioned transformation to transform coordinates from the first image space to the second image space. This “combined image space” may in turn be registered to the patient space using the techniques mentioned above.
Following registration, the two or more spaces are linked through the transformation calculations. Spaces that may be linked may include for example patient and image, image and image, or multiple images and patient. Once registered, the position and orientation of a tracked probe placed anywhere in the registered region may be located on, for example, a scan or set of scans of the region. Likewise, it may be possible to point to a location on one scan and have the matching location be displayed on another scan.
When performing an intervention, a tracking device may be used. Typically the tracking device if used may be connected to a computer system. Scans may also be loaded on to the computer system. The computer system display may take the form of a graphical representation of a probe or an instrument's position superimposed on to preoperative image data. Accordingly, it is possible to obtain information about the object being probed as well as the instrument's position and orientation relative to the object that is not immediately visible to the surgeon. The information displayed can also be accurately and quantitatively measured enabling the physician to carry out a preoperative plan more accurately.
In an image-guided intervention, it is desirable to plan the placement of a device or instrument in a precise pre-planned location (e.g., defining both its location in three dimensions (e.g., its x, y, z location) as well as its orientation (roll, pitch, yaw)). Because of the interdependence and coupling of orientation and translation, it is typically extremely difficult and tedious to manually align the instrument with the preplanned location in all 6 degrees of freedom (all translations and rotations) simultaneously even with computer feedback relaying the distance from the planned positions and orientations. As soon as some of the degrees of freedom are aligned, attempts to align subsequent rotations or translations cause the other degrees of freedom to fall out of alignment. While it is usually possible to converge on the correct alignment, it may take some time to do so. It is also not intuitive as to how to move the probe to easily achieve this alignment.
An additional concept in image-guided intervention is that of “dynamic referencing”. Dynamic referencing may account for any bulk motion of the anatomy or part thereof relative to a tracking device. This may entail attachment of additional position indicating elements to the anatomy, or other techniques. For example, in cranial surgery, position indicating elements that form the dynamic reference are often attached directly to the head or more typically to a clamp meant to immobilize the head. In prostate surgery, a special Foley catheter may be used to track the prostate with the use of a position indicating element embedded in the catheter (see U.S. Pat. No. 8,948,845 to Glossop et al., entitled “System, Methods, and Instrumentation for Image Guided Prostate Treatment,” which is hereby incorporated by reference herein in its entirety). In spine surgery, a dynamic reference attached (via a temporary clamp or screw) to the vertebral body undergoing therapy is used to account for respiratory motion, iatrogenic motion, as well as motion of the tracking device.
“Gating” may also be used to account for motion of the anatomy. Rather than continually compensating for motion through dynamic referencing, “gated measurements” are measurements that are only accepted at particular instants in time. Gating has been used in, for example, cardiac motion studies. Gating synchronizes a measured movement (e.g., heartbeat, respiration, or other motion) to the start of the measurement in order to eliminate the motion. Measurements are only accepted at specific instants. For example, gating during image guided intervention of the spine may mean that the position of a tracked instrument may be sampled briefly only during peak inspiration times of a respiratory cycle.
Both registration and use of an image guided intervention system in the presence of anatomical motion (such as that which occurs during normal respiration) is generally regarded as safer and more accurate if a dynamic reference device is attached prior to registration (and/or if gating is used). Instead of reporting the position and orientation of a position indicating element of a tracked instrument in the fixed coordinate system of the tracking device, the position and orientation of the position indicating element of the tracked instrument is reported relative to the dynamic reference's internal coordinate system. Any motion experienced mutually by both the dynamic reference and the tracked instrument is “cancelled out.”
With reference to FIG. 1, an organ 101 is depicted (e.g., a prostate gland, kidney, liver, thyroid, or other organ) containing a suspected tumor 102. Tumor 102 may be have been detected by an imaging modality such as MRI, multiparametric MRI, CT, PET, ultrasound, or by some other method. Once detected, it may be desirable to place a needle into tumor 102 for the purposes of biopsy, therapy, or delivering fiducials, for example.
The article by Pinto et al., entitled “Magnetic Resonance Imaging/Ultrasound Fusion Guided Prostate Biopsy Improves Cancer Detection Following Transrectal Ultrasound Biopsy and Correlates with Multiparametric Magnetic Resonance Imaging,” The Journal of Urology, Volume 186, Issue 4, 1281-1285, which is hereby incorporated by reference herein in its entirety, demonstrates the use of multiparametric MRI in the detection of prostate cancer. Once it is visualized on an imaging modality such as MRI, it may be annotated on the MRI scans. It may also, for example, be segmented so that its three dimensional boundaries are visible on the scans. The suspected cancer regions may be marked as single points, as indicated by asterisk (*) point 103. The spatial location, size, and/or orientation may also be modeled or notated and stored in a database or in reference to the images on which it was detected.
In some instances, an organ or region may be segmented or delineated so that its boundaries are apparent. This may assist a physician in understanding the boundaries of the organ. It may further assist in registering the position and orientation of the organ with subsequent images of the organ and, for example, enable it to be projected or fused into images obtained using another imaging modality. For instance, a three-dimensional graphic rendering representing a prostate gland that has been segmented from MRI may be fused with a real-time imaging modality such as ultrasound rather than the actual MRI images. The organ, in addition to critical structures within or around the organ such as important vessels, nerves, ducts, stones, bones, valves, nodes, and other regions of interest may be segmented.
As shown in FIG. 1, a number of needles 104a, 104b, 104c, and 105 are shown converging onto the tumor, specifically suspected cancer region 103. The needles may be positioned for the purposes of sampling tissue (e.g., for a biopsy) or delivering a treatment as mentioned previously. Both the position and orientation of the needles are important so that while needles 104a, 104b, and 104c may be acceptably placed, needle 105 may transect a structure 106 (e.g., such as the urethra) which may not be acceptable. Using the methods explained above, a physician would attempt to avoid this structure. For example, in a transperineal saturation biopsy of the prostate, a physician may use imaging to constantly monitor for a needle that will violate the urethra.
In a prior art depiction shown if FIG. 2, a needle 201 is equipped with an electromagnetic tracking sensor or position indicating element 202 that, when connected to a position sensor 203, enables its location and orientation in space to be detected. Position sensor 203 may determine the location of position indicating element 202 in a frame of reference 204 so that a transformation matrix “[T0]” may be reported that determines a translation and rotation to locate position indicating element 202 (and thus the tip of needle 201) in frame of reference 204. Similar devices have been disclosed previously for example in U.S. Pat. No. 6,785,571 to Glossop, entitled “Device and method for registering a position sensor in an anatomical body,” which is hereby incorporated by reference herein in its entirety.
A registration step may be performed to relate the position of the actual anatomy 206 in frame of reference 204 with the images 207 of the anatomy. This transformation is indicated as “[T1]” in FIG. 1. This enables a graphic display 209 of the needle on the pre-procedure images 207, which moves around as the needle 201 is moved. Needle 201 may then be placed into the lesions or suspected lesions 210 by observing the graphic display 209 of the needle while manipulating the actual needle 201. When the graphic display 209 of the needle is shown to be in the correct trajectory, needle 201 may be placed into the anatomy 206 and subsequently into lesion 210. There are numerous ways to perform this registration to obtain T1, some of which are referenced above.
In some implementations, an ultrasound, X-ray, or other live imaging modality may be used in conjunction with the pre-procedure images. In one implementation, an ultrasound transducer 211 may be equipped with a position indicating element 212 that indicates the position and orientation of transducer 211 relative to frame of reference 204, indicated here as transformation “T2.” If a calibration has been performed, the location and orientation of the scan plane 214 of transducer 211 is known as a fixed transformation “T3.” From this, points in the anatomy 206 on the scan plane 214 together with transformations T1, T2, and T3 can yield the location of these points on pre-procedure images 207, and it is possible to fuse the preoperative images with the live images. If the location of needle 201 is known through transformation T0, it too can be projected on the preoperative and intraoperative images.
Methods of ultrasound calibration to determine T3 are known in the art, some of which are summarized in the document to Gee et al., entitled “3D Ultrasound Probe Calibration Without A Position Sensor,” CUED/FINFENG/TR 488, September 2004 (Cambridge University, Department Of Engineering, Trumpington Street, Cambridge CB2 1PZ, United Kingdom), and in the document to Lindseth et al., entitled “Probe Calibration for Freehand 3-D Ultrasound,” each of which is hereby incorporated herein by reference in its entirety.
Templates or guides have been used in orthopedic surgery as well as a number of commercial manufacturing procedures that require cutting, drilling or assembly operations. Templates are guides that include guide elements such as holes and slits that are placed in a precise location over a work piece. Tools such as saws or drill are used to create holes or cuts in the work piece by first attaching the template to the work piece, and then placing the tools into the guide elements of the template and performing the cutting or drilling operation. The elements in the template are defined a priori so that if the work piece is properly aligned, the holes and cuts will be in the correct location.
In instances relating to orthopedic surgery, patient-specific templates have been employed. For example, U.S. Pat. No. 8,956,364 to Catanzarite et al., entitled “Patient-Specific Partial Knee Guides and Other Instruments” describes a cutting guide that differs from standard cutting guides used for total knee arthroplasty because it uses a template that is custom machined to the contours of the patient's bone to help align it. It is placed into the best matching position on the bone and, once in place, it may be used to guide the cuts in the bone directly or assist in mounting one or more cutting guides.
Medical templates are used exclusively for hard tissues, such as bone, since they can be aligned against hard immovable features on the bone. Unfortunately, soft tissues are not amenable to alignment using templates in the same way as bone, because templates cannot engage (and therefore be affixed to) soft tissues. These and other drawbacks exist.